Stage by stage2 min readUpdated 2026-11-09

How to generate and run the code

Generate code from the approved context.docx (Step 3), read it, then run it on your own computer (Step 4). Check the graphs and tables carefully before you write the manuscript, because the manuscript is built from them and can be created only once.

Code is generated from the approved concept, then run on your computer. You stay in control at both points.

Before you start

• An approved context.docx
• Python 3 installed with Add python.exe to PATH ticked
• Your dataset, or none for simulation work

Steps

1. 1

   Check the Ready check

   Code generation needs an approved context.docx, datasets found (or No dataset for simulation work) and Claude connected.

2. 2

   Set the options and press Generate code

   Choose the percentage of each dataset to use, whether to write quick self-checks for each stage, and whether to add comments that explain the method. The language is Python. The code is written into the code folder and is not run yet.

3. 3

   Read the code

   Browse the tiles (files generated, Python modules, main.py with a run-all screen, datasets wired in) and the file tree. Click a file to preview it, or press Open to open it in its program.

4. 4

   Set the run options

   In Code execution choose the percentage of data to use and a random seed (default 42).

5. 5

   Press Run the code

   Claude installs the required packages, runs the pipeline once for each dataset and checks the outputs. This takes about 25 minutes on a typical laptop.

6. 6

   Review the results

   Metric tiles for your method, a graph gallery (click to enlarge, Esc to close), result tables and parameter files. With several datasets, filter by dataset at the top.

Tips

Common mistakes

• Running code you have not read
• Using the wrong dataset path
• Ignoring warnings in the log

Questions

Where does the code run?

On your own computer, with Python 3. No other editor is required.

What computer is enough?

Windows 10 or 11 with at least an Intel i3-class processor and 16 GB of memory. If memory runs out during a run, the code is fixed rather than the step skipped.

Related articles

On the blog: How to share research code so others can reproduce your results (6 min read)

Last updated 2026-11-09. EasySeer is AI. It creates only an initial draft from the novel concept you provide, and it is designed to follow academic ethical guidelines.